KR19990045075A - Room temperature control and control room with feedforward / feedback control - Google Patents

Abstract

As a controller for heating, ventilation and air conditioning distribution systems, it has a feedforward control system and a feedback system. The controller includes a feedforward control system that generates control signals in accordance with control set values and verify characteristics of the system and suitably adjusts the set values according to measured changes in the verify characteristics. In particular, the controller is employed to control the temperature during the heating or cooling operation in the controlled space.

Description

Indoor temperature control device with feedforward / feedback control and its control method

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a control system, and more particularly, to an air conditioning system used in a fluid distribution device having heating, ventilation and air conditioning functions.

Background of the Invention [0002] Fluid distribution systems with a distribution system for heating, ventilation and air conditioning (HVAC) are generally well known and widely used in commercial buildings including, for example, apartments and office buildings. In addition, it can be seen that such a system is widely used in laboratory-type structures, and in the functioning of such a device, the HVAC system not only controls the room temperature of the building, but also includes a large number of laboratory smoke hoods Dispose of potentially toxic fumes if installed. In addition to the room temperature control function in the building, another important function to consider is the cleanliness of a clean room that manufactures integrated circuits and the like. In these two performance functions, the air pressure in the corresponding room must be controlled to be different from the air pressure in the adjacent room or the adjacent room. In the case of a clean room, the room must maintain a pressure higher than the surrounding space to prevent contaminants from entering the room. In the case of a laboratory, the corresponding room must maintain a lower pressure than the surrounding room so that it contains all the toxic gases in the room.

The present invention is mainly aimed at keeping the room temperature of the compartment at a predetermined value and enabling the compartment to maintain a predetermined differential air pressure with respect to the surrounding space. To this end, the HVAC system must be able to control the flow of air into and out of the room and all other air flows into and out of the room. The given temperature control requirements that must be maintained in the compartment result in complicated control problems that are difficult to solve easily.

On the other hand, an air volume variable (VAV) regulator is used to provide the above-described function performance control method. Such a regulator uses a combination of the feedforward control method and the feedback control method, There is a continuing need for an effective control device that provides easy functioning and cost effectiveness.

The main object of the present invention is to provide an improved heating / cooling indoor temperature control device having a feedforward control function and a feed-back control function, and a control method of the apparatus.

It is another object of the present invention to provide an improved controller with excellent performance, very easy performance and cost effectiveness.

SUMMARY OF THE INVENTION An object of the present invention is to provide an improved control system for a vehicle having an improved feedback control function for generating a control signal based on a control set value and an acknowledged characteristic value of the system and for adjusting the control set value appropriately based on a change measured on the confirmation characteristic value Controller.

It is yet another object of the present invention to provide an improved controller for uniquely employing energy conservation law and mass conservation law in the feedforward control function to determine the control setpoint employed in the feedforward control function.

Yet another object of the present invention is to provide an improved controller which is simple and superior in performance as a minimum computation time in the system identification operation by checking the characteristic values of the system using a general regression neural network (GRNN) .

It is still another object of the present invention to provide an improved control system capable of exhibiting excellent performance in various aspects by combining feedback processing and feedback processing to generate a control signal.

1 is a block diagram illustrating a controller and a control method according to the present invention;

Fig. 2 is a configuration diagram illustrating an embodiment of the feed forward control method of Fig. 1 employed to control a hot water heating coil and a water flow control valve;

Figure 3 - is a schematic diagram illustrating another embodiment of the feed forward control method of Figure 1 employed to control an air damper / actuator;

Fig. 4 is a configuration diagram showing an embodiment of the feed back method of Fig. 1

Fig. 5 is a configuration diagram showing another embodiment of the feed back method of Fig.

Figure 6 - Diagram showing normal flow rate versus normal control signal for a simulated valve with a power of 0.1 based on the sample flatness σ during validation using the GRNN method

Figure 7 - Diagram showing expected indoor heat load using multiple techniques

Figure 8 - Diagram showing steady flow versus steady control signal for simulated valve based on sample valve authority between 1 and 0.01 upon validation using the GRNN method

Figure 9 - Comparison table of simulated and predicted control signals for valves with powers between 1 and 0.01 when verified using the GRNN method

Figure 10 - Diagram showing normal feed flow rate versus coil efficiency for the simulated coils at the time of confirmation using the GRNN method

Figure 11 - Diagram showing steady flow rate versus measured normal control signal for damper upon confirmation using GRNN method

Figure 12 - Diagram showing the action on the pressure control sequence of the fume hood exhaust system

Figure 13 - Illustrative illustration of an interior differential pressure response comparing the performance of the first and second models

Figure 14 - Diagram showing cooling control action in the fume hood exhaust facility, in particular an example of heat release rate and flow rate versus time

Figure 15 - Cooling control action in a fume hood exhaust system, particularly a plot showing the ratio of heat release rate to flow rate versus time

Figure 16 - Diagram showing heating control action, in particular one example of heat generation rate and flow rate versus time, in a fume hood exhaust plant

Figure 17 - Diagram showing heating control action in the fog hood exhaust facility, in particular another example of the ratio of the rate of heat generation to the flow rate versus time

DESCRIPTION OF THE REFERENCE NUMERALS

10: Heating temperature control loop 12: Pressure control loop

14: Cooling temperature control loop 22: Feed forward control block

24 feed-forward control block 26 feed-

28, 29: physical system block 30: supply damper / actuator block

32: coil / valve actuator block 34: discharge damper / actuator block

36, 40: Online check block 38, 42: Control block

50: Switch

The present invention relates generally to a controller and method for determining control signals used in a method that combines a feed-forward control method and a feed-back control method to control temperature during heating with an HVAC system. Although the laboratory implementation is described in detail here, the proposed control method or result is useful in a clean room where the pressure is maintained above ambient to prevent contaminants from entering the compartment from the outside.

The controller according to the present invention uses a feedforward control means and a feed back control means in combination as shown in the configuration diagram of FIG. 1, and is divided into three distinct control loops, that is, denoted by reference numeral 10, to embody the present invention A pressure control loop 12 and a cooling temperature control loop 14, embodying the present invention. The three loops are functionally interconnected as shown by lines 16, 18 and 20, and all loops are preferably implemented in a processing means such as a microprocessor, although not shown.

The control sequence of the heating temperature control loop 10 embodying the present invention is shown in Figs. 16 and 17. Fig. The feed air fed into the test space in most air volume variants (VAV) has a constant temperature of about 55 ° F. Based on the design of a normal cooling load, the flow rate in terms of the volume of air supplied is selected so that the temperature of the designated room is always maintained between 70 and 75 degrees Fahrenheit. In order to maintain the differential pressure, it is necessary to increase the flow rate of the supply air such that the minimum value of the total laboratory discharge exceeds the feed flow rate due to the chassis opening of the fume hood. The flow rate of the fresh feed air having a constant temperature as 55 [deg.] F exceeds the required amount for cooling. Therefore, the room temperature falls below the set value. In this sequence, it is necessary to increase the temperature of the supply air so as to keep the room temperature at a set value while requiring the opening of the local disaster prevention valve. The coupling between room pressure and heat constraint is complex.

In the pressure control loop 12, the room pressure is controlled to a differential value instead of an absolute value. The differential value is defined as the difference between the reference space, that is, the adjacent region and its compartment itself. The goal in the laboratory is to maintain the differential pressure at a positive value within the range of about 0.005 to 0.05 w.c. This allows the room pressure to be kept below the pressure of the adjacent area under all operating conditions, thereby preventing air from leaking from the laboratory to the adjacent space. When applied to a clean room, the room pressure maintains a differential pressure higher than the adjacent space to prevent the room from entering the room.

Recently, there are three methods of controlling indoor pressure: direct control method, fluid tracking control method, and serial control method. These methods substantially regulate the flow of supply air to maintain the interior differential pressure. Therefore, a simple sequence is considered to evaluate the performance of the different control methods for pressure control. In the application for laboratory pressure regulation with a fume exhaust hood in the room, a change in the emission process in the fume hood requires adjustment of the amount of air supplied to maintain the differential pressure setpoint.

The pressure regulation sequence according to the present invention is shown in FIG. As shown in Figure 12, the discharge of fumes is raised to a maximum due to the opening of the hood chassis from the steady state. As a result, the pressure in the laboratory decreases and the pressure difference becomes larger. The pressure control loop 12 then senses the deviation of the actual pressure from the set point and opens the air flow to return to the set point.

Referring to the cooling temperature control loop 14, its control sequence represents temperature control as a result of the cooling demand. The rate of internal heat generation is the main disturbance force activating this sequence. The rate of internal heat generation can be increased by multiple redundancies due to other activities in the laboratory, such as high pressure cookers, ovens and occupancy periods. The room temperature rises when the internal generated heat suddenly increases. The only available cooling source is the air flow at 55 ° F. However, the supply amount of air can not be increased unless the discharge air is increased to maintain the differential pressure. However, laboratory emissions can not be increased because they disrupt the pressure in the laboratory. To avoid this problem, another source, i.e., all of the discharge devices, is opened to allow an increase in the amount of supply. As shown in FIGS. 14 and 15, by increasing the general discharge amount of the laboratory artificially, the indoor temperature and the indoor pressure can be maintained at the set values.

Each control loop 10, 12, 14 has a feedforward control block 22 and / or a feedforward control block 24, as shown in the configuration diagrams of FIGS. 2 and 3, respectively. 2 is for feedforward control of a heating coil, and the block diagram of FIG. 3 is for a heating temperature control loop, a feed forward control and a pressure control loop of a damper used in a cooling temperature control loop. Likewise, the fedbacker block 26 is identified in the control loops 10, 12, and 14, and the configuration of these fedback blocks is shown in FIG. 5 or FIG.

2 is provided with a coil and valve actuator 32 used for the operation of the controller and a physical system block 28 schematically showing the temperature sensor as described later. 3 also has a physical system block 29 schematically showing a pressure measuring means used for the operation of the controller and a flow rate measuring means. Likewise, the control loops 10, 12 and 14 are provided with a supply damper / actuator block 30 schematically showing a supply damper and an actuator coupled to an air supply duct in the room under control. The heating temperature control loop 10 also includes a circulating water heating coil for regulating the flow of the circulating water flowing through the heating coil and a coil / valve actuator 32 for schematically showing the circulating water valve. By placing in the air supply duct, the coil is employed to heat the air passing through the air supply duct. Finally, the cooling-air temperature control loop 14 includes a generic air discharge damper coupled to a common air discharge duct in the room being controlled and a common discharge damper / actuator block 34 illustratively showing the actuator. The general discharge duct is separated from the duct connected to the discharge duct or the laboratory mist discharge device located in the room, and discharges such as mist from the inside of the mist discharge hood together with air. The discharge operation of such a mist discharge hood is necessary to remove air from the room, and the controller compensates for such discharge as described below.

The operation of the feed back word block 26 will be described with reference to FIG. It employs a proportional-integral-derivative (PID) control method known to those of ordinary skill in the art of HVAC. The feed back controller utilizes the deviation between the set value and the measured variable according to the input, and the PID control method is used to reduce the process variable to the set value. The simple digital conversion of the control signal C _{s, m} from the PID can be developed starting from the minute continuous representation for the PID at the m-th sample time as follows.

Where S _t is the sample time, P _g , I _g and D _g are proportional gain, integral gain and differential gain, respectively. The first term in the right direction of the equation represents a constant offset value. The second term is the proportional action factor, the third term is the integrative action factor, and the last term is the differential action factor.

A similar expression can be written for the (m-1) -th sample as follows.

By subtracting the second equation from the first equation, we can get something that is easy to implement in a digital controller.

Describing the feedforward control method, a physical model is used to determine the control variables, i.e., the feed air flow rate, the feed air temperature, and the setpoint for a typical discharge damper. The specific control variable is chosen based on the application. Application is defined as a sequence of events initiated by disturbances of pressure and temperature in the laboratory that require the controller's response to change the process variable, ie, the state of the control variable. For example, if the total emissions of the laboratory suddenly increase due to the opening of the mist discharge hood chassis, the room pressure decreases. Therefore, the flow rate of the feed air must be increased to maintain the room pressure at the set value. In this example, the feed air flow rate is a control variable, whereas the total discharge flow rate of the laboratory or the differential pressure across the laboratory is one of the processing parameters dependent on the measured value.

The second stage of the feedforward controller includes the first phase and the generation of control signals based on the setpoints determined by the characteristics of the HVAC plant. Two types of equipment are commonly known as HVAC systems in air volume variable (VAV) laboratories. These include valves or dampers that limit the flow of water or air through water / air heating coils, which typically heat laboratory supply air. The characteristics of each component are correlated with the output as input variables and control signals.

2, the feedforward control block (controller) 22 is connected to the online confirmation block 36 and the control block 38. Similarly, the feedforward control block (controller) 24 shown in FIG. A confirmation block 40 and a control block 42 are provided. The verification blocks 36 and 40 receive and update processing characteristics according to the input control signal and measurement variable of the processing device. The verification blocks 36 and 40 periodically send the update characteristics to their respective control blocks 38 and 42 for control actions.

In such a situation, it should be noted that the feedforward controller has a feed-back mechanism in common for compensating for changes in system characteristics. However, the feedback mechanism described above differs from the fed back method in that the measured processing parameters are compared with set values to generate an error signal and the output signal is substantially a function of this error signal. Even process variables and system disturbances in the pay-forward verification process are measured when they are cost effective and convenient. The feedforward control blocks 38 and 42 operate based on receipt of the setpoint signal and supply a control signal based on the acknowledgment characteristic for that process. The essence of the feed-forward control method is to output a control signal as a response to a change in a set value or a measured variable of the processing apparatus. Since the feed forward control does not require an error signal to generate a control signal, it responds faster than the feed back control.

The verification process identifies the characteristics of the system over the entire operating range and creates a powerful controller. If the verification system can fully understand the characteristics of the entire system, it does not need a feedback controller. However, complete characterization can not be realized without incurring high costs due to errors in data, noise, and accuracy of data.

For each unit of the control facility in the HAVC system of the VAV laboratory, the feedforward controller can generate a control signal according to the set value change of the process variable. To understand how control signals can be generated, we need physical processing coupled with each component.

A physical processing device that raises the temperature of the room includes two components: a valve / actuator assembly and a heating coil. The VAV laboratory is equipped with heating coils, valve / actuator and damper actuators to satisfy the laboratory pressure and temperature conditions. The characteristics of the valve actuator are similar to those of a damper / actuator used to regulate the air flow rate in a HAVC air distribution system. Therefore, the processing operation described with respect to the valve here is equally applicable to the damper / actuator. An example of the verification of all HAVC components in the VAV laboratory by selecting an example of the heating operation is given.

The rate of flow of water through the valve depends on the opening of the valve and the authority a. Privilege is defined as the rate of pressure drop across the valve against the pressure drop across the entire circuit when the valve is fully open or when each valve is equal to (Equation 5d)

It is common in the art to express valve characteristics in terms of power, valve opening rate and maximum flow rate (ASHRAE 1992).

Indeed, in a single circuit system, the pressure drop of the circuit will be small compared to a valve whose power a is close to 1.0. However, in a system equipped with multiple circuits, as the distance between the pump and the coil increases, the pressure loss of the main portion becomes larger than that of the side portion. As a result, the magnitude of the authority varies with the rate of pressure loss as shown in the equation of authority. The power of all circuits is time-dependent because the flow in each circuit varies with time. The authority of the valve can be determined by using the basic relationship between the pressure drop and the flow rate by design or by measuring the static pressure at the pump outlet and valve inlet under flow conditions at the time of design and calculating the power at all times.

As shown in FIG. 2, a control signal _Cs is generated in accordance with the heating demand, and is sent to the valve / actuator 32 so as to open and close the valve. The heating coil is equipped with physical input values such as water flow rate, air flow rate and incoming and incoming air temperatures. The output of the coil is the water temperature and the exhaust air temperature. The outlet water temperature is not directly related to the thermal energy of the supply air, so it is not considered in the verification process. Instead, the input water temperature T _{f, i} and the variable R, which are independent of the inlet temperature and the exhaust temperature T _{a, i} , T _{a, o} , respectively, are used. T _{f, i} and T _{a, i} are known or measured as a given constant of the system, such as user input parameters, and are input to the controller. Also, the variable R, which can not be seen as the coil efficiency, is the amount of heat supplied. R can be expressed by the following equation (5e).

R = (T _{a, o} -T _{a, i} ) / (T _{f, i} -T _{a, i} )

The physical processing described above is related to the system processing variables as a function of the control input. The above physical processing needs to be reversed when used in a feedforward controller to generate a predetermined control signal that adjusts the valve to a predetermined position in accordance with the water flow rate setting.

This control outline can be explained with reference to FIG. The physical heat treatment procedure described above is reversed in the feed forward block shown in FIG. The feedforward block includes a heating coil T _{a, | sp} Lt; / RTI > The on-line verification unit normalizes and inverts the characteristic to generate a predetermined control signal. The characteristic of the heating coil is determined by first setting a predetermined discharge air temperature set value T _{a, | sp} And a given feed air flow rate Lt; RTI ID = 0.0 > Is used in the control processing process. Upon knowing the demand and authority a of the water flow rate, C _s .

Control signal C _s The variables observed from the physical system are periodically collected and used to update the characteristics of the coil and the characteristics of the valve by means of a separate verification unit shown in FIG. Observed variables include T _{a, o} , T _{a, i} , And . However, instead of the expensive means for measuring the water flow rate, T _{f, o} And using the following energy balance equation Can be calculated.

Where K is a constant determined experimentally and expressed as the ratio of mass to mass of air and water.

here ρ _a The air density, ρ _f Is the fluid density, C _a The air capacitance and C _f Is the fluid capacitance. The above water flow rate ( ) Equation is preferred as a method of calculating the flow rate of water flowing through a localized heating coil, taking into account cost and practicality, as opposed to measuring direct flow velocity. The HVAC control system tends to consider the air flow rate through the heating coil as well as the exhaust air temperature for control purposes. The above values are updated every second or more. In addition, the incoming air temperature and incoming water temperature of the heating coils can often be used from the central air control unit and the cooling unit. By adding the water temperature sensor in this way, the water flow rate of the heating coil can be controlled by the water flow rate ( ) ≪ / RTI > equations. Because the flow rate sensor is more expensive than the temperature sensor, such a cost difference is very cost-effective because it is very large considering the large number of heating coils that can be installed in each area of the building. In addition, the band attachment temperature sensor at the time of product update can be installed outside the pipe in order not to disturb the business in cost. On the other hand, most flow rate sensors need to be embedded in pipes, which interferes with system operation.

The following additional factors encourage the use of temperature sensors. First, the water flow rate equation is used for verification purposes only. Therefore, it does not require dynamic data to solve the water flow rate equation. Instead, it only needs periodic steady state data, which makes it difficult to obtain a given velocity of one or more preferred samples per second. Second, the dominant relationship between water flow velocity and air flow velocity and the differential temperature of air and water across the heating coil are important in estimating the water flow velocity of the heating coil. Therefore, the absolute accuracy of each measure is not so important. Finally, the purpose of feedback controllers combined with the feed-forward approach and feed-back approach is to compensate for the inaccuracies in the validation process involving measurement errors.

Fig. 2 is a graph showing the relationship between the outlet temperature set value of the heating coil < RTI ID = 0.0 > T _{a, o | sp} The need for prediction of In fact, the characteristic reversal of the heating coil generates a water flow rate setting through the valve. By knowing the valve authority and the water flow rate setpoint, the controller can generate the control signal to the valve side.

The action described for the valve is similar for a damper, and for a damper is shown in Fig. In the case of a damper, the control signal is generated by the airflow rate requirement. The air flow rate set point is first determined and the damper authority is used by the feedforward block to generate the control signal accordingly.

According to an important aspect of the present invention, there is a need for a method for determining setpoints for feed air flow rate, feed air temperature and general exhaust air flow rate. The feed air flow rate setting is associated with a pressure loop for the safety of the laboratory. The supply air temperature set value is determined when the indoor temperature falls below the set value and needs to be heated. The general discharge device is opened when the room temperature rises and exceeds the set value. In all cases, a physical model is used to calculate setpoints.

Fixed mass resin and infiltration equations can be used to solve feed rate setpoints to determine feedrate setpoints. When the fixed mass resin is described in the set-point term, it is equal to (Equation 5h).

The infiltration relation, which is the amount of air flowing into the room from other supply pipes, is shown in the following equation (5i).

Laboratory differential pressure values ΔP _{| sp} Is defined as follows.

? P _{| sp} = _{pref | sp} -p _{| sp}

There are nine variables in the mass water lipid formulation, which are temperature, flow rate, and pressure for three air streams: supply, infiltration, and laboratory discharge. The setting values for the infiltration temperature and pressure of the room are known. The set value is The volume of infiltrated air is Wow ΔP _{| sp} It is known from the equation. Similarly, the supply air pressure P _{s | sp} , Room pressure P _{| sp} Then, T _{| sp} Is given from the design data. Unknown is the laboratory supply airflow rate , Laboratory full set point The air release temperature set point T _{s | sp} . The total emissions in the laboratory are the sum of the general emissions and the emissions from the fume hood, as given in (5k).

In the VAV laboratory, the emissions setting of the fume exhaust hood is the known amount of the position of each fume exhaust hood chassis. Therefore, it is possible to know the general discharge setting amount by determining the discharge amount setting value of the whole laboratory.

The following energy invariant equations are used directly to add the four preceding equations to solve the supply air release temperature or the general emission setting. The energy invariant equation is as follows.

here C _f Is a unit conversion factor.

It should be noted that when the supply air release temperature setpoint has been determined, the normal emissions may always be known amounts or vice versa. There is a need to determine a predetermined supply air release temperature when the chassis of the mist discharge hood is opened to suddenly increase the discharge amount. The increase in the emission of the exhaust means requires more supply air to maintain the room pressure differential. However, if the amount of feed air, typically 55 ° F, exceeds the amount required to offset the cooling load to maintain the room temperature at 70 ° F, the room is subcooled. In order to prevent the subcooling of the room, the supply air must be heated and the heating coil adjusted to achieve the predetermined supply air temperature setpoint.

Typical emissions are needed when the fume exhaust hood is closed and the amount of heat generated on the inside increases due to the operation of the experimental or experimental equipment. In such a situation, the room needs to be further cooled. However, additional cooling by increasing the volume of the 55 ° F supply air results in an upward pressure balance adjustment. As a result, the general discharge damper is opened so that more supply air can be supplied for additional cooling. The controller shall control and determine the general discharge amount and the supply air amount in order to maintain the indoor pressure set point and the room temperature set point. In this case, of course, the temperature of the feed air is fixed at 55 ° F. The normal discharge damper is always closed when a temperature rise is required, Quot; 0 ".

Therefore, the use of five preceding equations yields setpoints for feed air flow rate, feed air temperature or feed air flow rate and normal exhaust flow rate, according to the control sequence. In the last equation, the energy invariant equation, q _load Needs to be determined in order to obtain the set value. The instantaneous load is almost proportional to the first order differential of temperature over time. This is the internal energy storage period, assuming that the laboratory air mass remains constant.

q _{load | tr} =? C _v dT / dt

The room temperature T can be measured directly by installing a temperature sensor in the exhaust duct of the room instead of following the routine installation practices of wall thermostats. In many laboratories, the fume exhaust hoods piped together in common between the two exhaust streams and the exhaust pipe from the laboratory provide a good place for duct temperature sensors. Due to the high air conditioning conditions, the air in the laboratory is well mixed and therefore the exhaust air temperature is a good representation of the room temperature T. In some situations, however, the installation of a duct temperature sensor may not be feasible because the voltage supplied to the sensor may respond to volatile haze. In such a situation, a thermostat sensor for wall mounting can still be used and the room temperature can be predicted by using a temporary indoor air temperature sensor, as described below, by simplifying the following equation:

This equation is based on the thermostat temperature T _st Side wall temperature and the indoor air temperature. It is necessary to combine the wall temperature with the thermostat temperature since the energy radiation of the wall heats or cools the thermostated wall. In most laboratories, the temperature of the wall is very close to the temperature of the space, since both the laboratory and the space adjacent to the laboratory maintain the same temperature in the inner region. As a result, the above equation can be simplified as follows.

Only thermostat adjustment constant C2 _st The temperature sensor may be temporarily disposed in the exhaust pipe or disposed at a suitable position in the room during the trial operation, C2 _st To determine the thermostat temperature T _st And the indoor air temperature T from the temporary position data and the fixed trend data. Once the thermostat constant is adjusted, the temperature sensor is removed from the temporary position. Instead, if possible, the sensor for measuring indoor air temperature can only be placed in a common exhaust duct for laboratory air. Sensors installed in common exhaust pipes can not be used continuously instead of thermostats because the common exhaust damper is completely closed and the sensor is not exposed to indoor air. On the other hand, by installing a sensor in a common exhaust pipe, when the flow of the common exhaust pipe is important, the adjustment constant C2 _st The adjustment process can be automated to update the value of < RTI ID = 0.0 >

When the room temperature is fixed, the following energy equation can be used to determine the total cooling load. This relates to the laboratory overall discharge rate, room temperature and feed flow rate at a pre-load time t-1. It is assumed that the air density is ideally constant for supply, discharge and infiltrated air.

Total laboratory emissions are expressed as the sum of the general emissions and the emissions of the fume hood exhaust.

In the equations (5p) and (5q) above, in order to prevent oscillation when the indoor load is estimated, Is used instead of the actual infiltration flow rate. The instantaneous value of? P is the infiltration flow rate And the room temperature T. As a result, the calculated indoor cooling load is oscillated.

Instantaneous value on the calculated load ΔP Wow A simple control method is selected to see the effect of the simulation. By increasing the indoor heat generation rate from the fixed value of 82.50 Btu / min to 412.50 Btu / min, the response pressure and the response temperature of the room side were obtained. Because the room temperature is increased with the higher incidence of the inside, more cooling is required on the indoor side.

Additional cooling may be provided by increasing the feed air flow rate to only 55 [deg.] F. However, prior to increasing the feed air flow rate, the total emissions in the laboratory must be increased to maintain the room pressure differential. Next, we will need to increase our general emissions. The use of the infiltration flow rate setpoint in predicting the load is here to determine the feed air flow rate and feed air temperature or the general exhaust flow rate that the object needs to obtain the room pressure differential setpoint and temperature setpoint I can tell by work. In effect, the controller drives a supply damper and a common exhaust damper to maintain an indoor differential pressure of 0.05 w.c and an indoor temperature of 70 ° F. The controller first calculates the setpoints for the flow rate of the feed air and the flow rate of the common exhaust air at fixed conditions before and after increasing the internal heat release rate.

Based on the set air flow rate, the controller determines the position of the damper according to the relationship between the flow rate through the damper and the damper position. The purpose of using a simple simulation is to illustrate that instantaneous values of pressure and temperature have a momentary impact on the infiltration rate. The resulting effect is that the expected load will follow a momentary change in the infiltration rate, which is vibrational. The instantaneous load under fixed conditions is the actual laboratory total discharge The above-mentioned q _{load | ss} Is determined by applying the equation. On the contrary, the estimated fixed load q _{load | ss} Will follow closely and closely with the actual load during an instant, and will match the simulated experimental load including the effects of fixed generation heat and walls. Stationary q _{load | ss} The ΔP Assuming a set point for To calculate q _{load | ss} Use equations. As a result, 0.05 < RTI ID = 0.0 > q _{load | ss} Is used in equation (5p). It was found that for the selected control sequence, the difference as determined by (equation (5p)) was about 41 cfm, q _{load | ss} To about 43 Btu / min.

Based on the observed behavior, the expected fixed load is chosen to be used in the simulation instead of the instantaneous load. In addition, the controller does not have to follow the actual instantaneous room load that causes the damper to oscillate. Use of the predicted load based on the set point provides a stable control state.

When it is necessary to cool the room, the storage term and the fixed load term are used to determine the normal discharge flow rate setpoint and supply airflow rate setpoint, q _{load | ss} Is added. However, in the case of heating only, the storage term is neglected in the operation of calculating the supply air temperature set point.

In the verification step, the output component is generated based on the information related to input, output, and various variable amounts using the characteristics of the verification element. There are two components to be checked: heating coil and valve / damper. However, since the physical characteristics of the control processing unit will be reversed as described above, the confirmation processing unit should emphasize the relationship between the input and the output of the inverted physical processing unit. For example, for a heating coil, its input is a variable R with no magnitude, T _{a, i} , The inlet air temperature of the heating coil and the inlet water temperature T _{f, i} to be. The output of the heating coil is the flow rate of the water flowing through the heating coil to be.

Likewise, referring to FIG. 2, in the confirmation operation of the reversed physical processing action on the damper and the valve, two input signals include a flow rate and authority and a control signal as one output signal. The damper or valve is a substantially variable fluid resistance device. Representation of the characteristics of a similar fluid and its performance can be expressed in terms of confirmation variables and thus can be expressed in the same model.

The Generalized Regenerative Neural Network (GRNN) is selected to identify the heating coils and valves due to simplicity, robustness and excellent performance in system identification. Unlike the conventional neural network, only a minimum computation time is required to effectively detect system characteristics. The following is a simple calculation example of a GRNN to illustrate its implementation means only upon confirmation of a component.

Input to GRNN is a series of data that can contain multiple sizes. Sample values of input vector X and input scalar Y X _i Wow Y _i A predicted value for a given average value Y at any given value X can be found using all sample values in the following relationship:

Here, a scalar function representing the distance of the Creed distance from a given value to a known position D _i ² Is given as follows.

D _i ² = (X - X _i ) ^T (X - X _i )

And σ is a single flat parameter of GRNN. The above equation is a practical GRNN method. It is assumed that the flat parameter σ is non-Gaussian except for the possibility that the expected density may vary widely between known irregular points for small values. When sigma is very large, the regression surface becomes very gentle. The holdout method (by Specht in 1990) is used to calculate the flat parameter value σ.

Also, the processing means of the GRNN for the characteristics of the heating coil and the valve / damper provides an effect beyond the conventional verification method. For a typical regression method, the operator must either enter a known equation or search for the equation best suited for the exhaust. Code in non-linear conditions is intensive and can be prohibited for efficient online use. On the other hand, GRNN uses significantly simpler code without requiring user input in the form of a function. Furthermore, the GRNN algorithm can be embodied in a neural network hardware processor. This eliminates the need for software input during site installation, thus eliminating a significant amount of software development activity.

The input vector X of the heating coil is a variable R with no magnitude While the output Y includes the flow rate of water flowing through the heating coil to be. The valve authority a and , The valve GRNN will produce the output Cs of the required valve control signal. In the damper / actuator for flow control, the input and output variables are the same as for the valve.

According to another important aspect of the controller, the coil characteristic and the valve characteristic are generated using the above model, and then GRNN is used to confirm the characteristics. Physical variables are normalized first. In addition to R (Equation 5e) and authority a ranging from 0 to 1, other used normalized variables

to be.

In this embodiment, c _smax , And Are 1.0, 2500 cfm (1180 L / s) and 1.0 gpm (0.0631 L / s), respectively. The load and the given value Using the value of R required to satisfy The value of the control signal < RTI ID = 0.0 > nC _s Lt; / RTI > may be used in the valve model according to a given authority to generate the desired pressure. Using the normalized parameters and the model described above, the characteristic data of the coil and valve of Table 1 are generated.

The GRNN method can best be described by using a regression example of valve data for constant authority. For example, if the authority is chosen to be 0.1, then a non-linear relationship is established between normalized control signaling normalized flows as shown in FIG. There is only one input for invariant powers and the vector X with the scalar function equation above is a scalar continuous averaged flow rate . Functions in scalar function equations D _i ² silver X _i Lt; / RTI > is the ith sample of the series of values. The GRNN equation for Y (X) D _i ² Use of verification data nC _s As the ith sample of Y _i Can be solved.

Valve simulation parameters _{(λ = 0.00001; W f =} 1; K cd = 0.08641 (64.89); K 0 = 0.042 (31.54);) Authority maximum gpm (L / s) 1.00 -0.086 (-64.58) 3.00 (0.1893) 0.70 -0.034 (-25.53) 2.50 (0.1577) 0.50 0.037 (27.78) 2.12 (0.1337) 0.20 0.407 (305.63) 1.34 (0.0845) 0.10 1.02 (765.97) 0.95 (0.0599) 0.05 2.25 (1689.64) 0.67 (0.0423) 0.01 12.13 (9109.02) 0.30 (0.0189)

Simulation of coil characteristics and valve characteristics, as well as GRNN, is performed using the Engineering Equation Solver (Klein and Alvarado, 1997), which is specifically incorporated herein by reference. The simulated data in FIG. 6 is shown in bold lines, while each point is generated by use of the GRNN equation for various flat parameter (?) Values. Although small values of σ do not appear to represent good data, we should avoid excessive fixation by choosing very small σ. The simulated data is increased from 0.0 to 1.0 by 0.1 nC _s And 0.05, 0.15, and 0.25 nC _s And 14 samples obtained by the above-mentioned method.

The holdout method (Specht 1990), which is specially combined here for reference, is used to calculate the optimal σ and is found to be 0.01 here. The higher value of the sigma selection effect is clearly shown in Fig. Unlike σ, which is a small value, σ of 0.5, which is a large value, it differs greatly from the input and it is found to be almost straight like a straight line. The GRNN method attempts to estimate all samples and is not gentle between points. When σ is 0.01, the average error between the expected signal and the simulated signal was found to be 2.62% and the maximum error of 14% nC _s ) For the lowest value of the control signal. Also, since the control signal is very sensitive to the average flow rate, A minor error was found in

The relative error of the higher side of the valve curve is very small compared to the lower end due to the higher control signal magnitude at this end. Therefore, the size of the sample and the choice of the sample are very important parameters depending on the flat parameter σ. In fact, nC _s The error between the simulated control signal and the expected control signal for a particular sample was reduced from 14% to less than 1%, while the average error fell from 2.62% to 1.31%. For verification of damper / valve characteristics, only at most 200 samples are required to cover the entire operating range. This means that the power can be varied between increments of 0.10 from 0.001, 0.01, 0.05 and 0.1 to 1, while the control signals can be varied from 0.05, 0.075, 0.01, 0.15, 0.20, 0.25, 0.30, 0.35 and 0.40 to 1.0 Lt; RTI ID = 0.0 > of < / RTI > In all technical states, the local controller can process 200 sample values easily and quickly. In practice, however, the number of points in the total area covering the actual operating range is very small, preferably less than 100.

Valve authority ranges between 0.5 and 0.1 were chosen to test the GRNN method. Again, the holdout method is used to determine the optimal flat parameter σ, which is currently 0.05, which produces a squared error of 0.189 over the confirmation data size of 30 samples. The verification dataset has a permission value of 0.10, 0.30 and 0.50 and an equal interval between 0.10 and 1.0 nC _s . The test data set is changed from 0.05 to 0.95 incrementally to 0.10 and also has intermediate authority of 0.20 and 0.40. An average error of about 3.0% is low compared to the range of the data set. Where the curve becomes very steep at normal speed, some errors above the mean have been found for the higher value control signals.

The operating range of the valve and the damper symbolizes such a controlled object. Therefore, the use of GRNN to represent characteristics using a small data set represents the feasibility of real-time controllers based on online. In practical applications, operating characteristics over the entire operating range may be revealed during commissioning by varying the damper opening area. Once determined, the operating characteristics are stored in the feedforward controller and control signals are generated based on the stored data using GRNN. The time and effort required to adjust the feedback loop is reduced because the error signal for the feedback loop always has a low value. Reduced commissioning costs and time and improved system performance are two key factors in promoting the combination of feedforward controllers and feedback controllers for building HVAC distribution systems.

The measurement data acquired during the commissioning process is only used to initiate the confirmation procedure. Since the system is operational and collects a lot of operational data, the confirmation data is updated accordingly. The essence of the combination of feedforward and feedback is to use the feedforward block to generate the control signal of the summation while using the feedback block to process the detail. In fact, the feedforward block also has a feedback mechanism to update the confirmation block. However, the checking process is separated from the control process in consideration of easy implementation and economical efficiency.

Another way to implement GRNN in a controller is to generate the characteristics using the simulated data. The characteristics can be stored and updated as actual data becomes available and replaces simulated data.

Figure 8 shows verification data and test data over the entire operating range of the valve. They were obtained by simulating a control signal changed between 0.1 and 1.0 for each authority in a confirmation set where the authority varied from 0.01 to 1.0. Additional samples were also superimposed from the test set to the confirm set in the low-valued rights and control signals. A total of 60 samples were used in the confirmation set and a total of 150 samples were used in the test set. To optimize the σ value, a holdout method using small data sets with permissions of 0.01, 0.10, 0.25, 0.50 and 1.0 was used. A small data set with a sparse value still makes a good choice with a sigma of 0.01 for the data set shown in FIG.

A comparison table of the simulated control signal and the predicted control signal is shown in Fig. Errors higher than the average error are also generated for a larger control signal as well as for lower rights. A large error for a particular sample can be significantly reduced by including the sample in the confirmation set. This can easily be achieved in a real-time controller by comparing the control signal generated by the valve and the damper with the control signal generated by the feedforward control signal. If the difference between the feedforward control signal and the total control signal increases above a predetermined fixed threshold, the control signal and its corresponding average flow rate and authority may be returned back into the acknowledgment set.

Finally, the GRNN method is used to verify the characteristics of the heating coil. Referring to FIG. 2, the GRNN method needs to anticipate a given water flow rate through the heating coil for a given R and air flow rate. The arbitrarily selected steady feed air flow rate value And normal flow velocity for R The The energy balance equation for K, the mass capacitance equation for K, and nC _s , And Can be calculated by using a normalization equation to obtain a normalization equation. Some of the simulated data is used for verification purposes, while the rest are left to test the GRNN algorithm. The test sample is deliberately chosen over the entire operating range. 9 shows verification data and test data.

An average error of 2.6% was found between the expected flow rate and the simulated average flow rate. In the valve, the heating coil shown in Fig. 10 appears randomly, as opposed to the definite pattern being obvious. The flow speed of the heating coils can be predicted with high accuracy at the same height as the distribution of the rare and random input data.

In addition to the simulated data, the measurement characteristics of the damper are used to test the GRNN. Two sources are used to obtain the measured values: 1) testing the data taken to adjust the damper operation; 2) activating the damper operation at the working position using the building automation system (BAS). In the first case, a damper curve for three damper powers was experimentally generated as shown in Fig.

The test sensors used to obtain the data are similar to those commonly used in building automation systems. For a given control signal, the flow rate through the damper is known, which is normalized using the normalization equation. As shown in FIG. 11, the GRNN is identified using the control signal and flow rate and power measurements, while the direct points on the power curve are used to test the GRNN.

Compared with the simulated data, the measurement curves of FIG. 11 are more random as expected. At low flow rates, three power curves converge to one point, indicating that it is difficult to measure the flow rate when the damper is fully open. Increasing the control signal at the fast flow rate and low authority value does not increase the flow rate. GRNN estimated the measurements with an average accuracy of 4.30%, which is good considering the errors associated with the measurement and data acquisition systems. The holdout method is used to determine the optimal flat parameter with a sigma of 0.066. As can be seen in FIG. 11, the error increases at a higher flow rate as the authorization curve becomes very sensitive. The range of test data for GRNN was chosen over the mean operating range of the damper between 10% and 100% of the flow rate.

The authority for the damper in the working position remained unchanged to 7% during the data collection. For the same flow rate, the control signal of the damper was varied widely at both high and low flow rates. The GRNN output is tested for each sample observation being used in the confirmation data. The preprocessing of the raw measured values is not used until they are sent to the GRNN to verify the data. A pre-processing filter may be used on the measurement to reduce uncertainty about the measurement.

The accuracy of the GRNN is estimated to be less than 6% when the control signal is expected. In addition, the linear regression of the valve characteristic shows an average error of 7%. The nature of the GRNN method lies in its ability to predict linear and nonlinear characteristics without any user input for fixed flat parameters. In the case of a regression tool, which is the primary user input for erasing a form, the regression often requires limiting the actual online implementation of regression for confirmation. Thus, the results demonstrate that the execution of GRNN outweighs the performance of linear regression.

The feed-forward-feedback combination method requires little adjustment because it allows most of the control signals to be provided from the feedforward block, thereby corresponding to only a small fixed error in the pie back block. Unlike the feed back loop, the feed forward loop only operates based on the value of the set position and does not require a measurement of these variables. As a result, the feed forward signal can improve the control speed when tracking changes in the set point. Most conventional methods employing feedback are a typical approach using a proportional-integral-integral (PID) algorithm and are suitable for a coupled approach.

Local controllers can be used in the implementation of the devices shown in Figures 1 to 4, many of which are found in medium to large buildings, and have sufficient memory and processing power for economy. A control method that is substantially simple, easy to implement, low in cost, and substantially improved performance can be provided by combining the feed-forward algorithm and the feed-back algorithm. Provides enhancement across the PID controller that reacts to the control being affected by the heating coil signal and the dynamic response of the valve signal. The static characteristics of these devices are stored and updated in the feedforward block as described above.

The combination of the feed forward block and the feed back block is preferably performed in one of two ways. As a first choice, a simple switch 50 may be used to set the control signal to zero from the PID algorithm whenever the set value is detected, as shown in FIG. This is confirmed as Model 1. Only the feedforward block generates a control signal when the set point is changed. The PID output is added only when the setpoint has not changed, indicating that the system is in a fixed state. This combining method is based on the fact that the feed back block can be accounted for only in the static state error that will not be detected by the open feed forward block. It is reasonable to expect a relatively small fixed error due to the uncertainty coming from the verification system, measurement and controller.

As a second approach, there is model 2 shown in FIG. 5, where the output of the network controller is the sum of the feedforward output addition, the integral part of the PID output, and the subtraction part of the proportional part of the PID output. The logic employed here is to subtract the proportional output so that feed-back remains inactive for all changes of the set position. The Piveback provides only integral and differential action so that the controller can respond to the setpoint change by the feedforward block.

Both of the two coupled models are simulated and compared to each other using a simple pressure control sequence to illustrate the response. As shown in FIG. 13, both models performed satisfactorily, although Model 1 performed better than Model 2 during undershoot and response times. The trend during the decrease of the flow rate is exactly the opposite of the trend during the increase of the flow rate. The performance of the controller is greatly improved over shorter sample periods. Sample time is a function of the processing speed and communication speed of the controller and is often costly. The controller preferably has a sample time of 1/10 second or 10 samples per second.

The GRNN method is effective in identifying the characteristics of the HVAC components for the sequence used for control. The ability of GRNN is evident in its ability to apply both linear and nonlinear relationships using both observations of simulated and observed samples. However, unlike typical return dynamics, prior knowledge of the relationship of the equation terms is not required to implement GRNN. The essence of GRNN is to be able to embed it in a neural network structure that enables hardware implementation. The flatness parameter is only a variable that needs to be selected and can be determined using a holdout method or other method.

The GRNN method is a useful tool for characterizing the static performance of HVAC components for use in feedforward blocks coupled with feedback controllers because small datasets are required for the characteristics of local HVAC control components: valves, dampers, and heating coils. / RTI > Multiple outputs can be processed by GRNN, even though the output Y is treated as a scalar in this description.

It is reasonable to estimate the error of 6% for the coil and valve characteristics by the GRNN method depending on the use of the measurement data. Therefore, the control signal can be generated with an average accuracy of 8.8%. The feedback controller is suitable for generating a control signal to eliminate residual error of 10% or less. The feed-back controller also requires minimal adjustment because the error range is expected to be within a fixed low range.

The combined model 1 shown in FIG. 4 using the PID controller under the fixed state showed better performance for the room pressure control than the model 2 shown in FIG.

It will be appreciated that a controller with substantial control capability as described above, which is simple, easy to implement, low in cost, and substantially improved in performance by the combination of a feed-forward algorithm and a feed-back algorithm, is illustrated and shown.

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art, It is possible without departing from the spirit and scope of the present invention as determined by the claims.

Claims (26)

(HVAC) system having a supply pipe for supplying air to the room and an exhaust pipe for discharging the room air, and the HVAC system has an element for controlling the flow of air supplied to the room A room temperature regulator for controlling a temperature of a room having at least one adjacent room in a building having a room having at least one additional exhaust means independent of the HVAC system, A feed forward control signal is generated in accordance with confirmation of the predetermined temperature set value and the predetermined flow set value of the exhaust pipe, the flow flowing into the indoor side and the flow discharged from the indoor side, the flow set value of the general exhaust pipe, Feed forward means, A feedback means for generating a feedback signal according to the measured system parameter, and And means for combining the feed forward control signal and the feed back control signal to control the local means. The apparatus of claim 1, wherein the air flow control element comprises a damper / actuator for supply, a damper / actuator for general exhaust, a heating coil, and a valve / actuator. 3. The method of claim 2, wherein the checking characteristic comprises at least one of an air flow rate flowing into the room, an air flow rate set value flowing into the room, a feedforward control signal, an authority of a supply damper / actuator, The flow rate of the exhaust air flowing out of the room, and the flow rate set value of the exhaust air flowing out from the room. 2. The indoor temperature control device according to claim 1, wherein the air flow regulating element includes a valve for controlling the flow of water flowing through the heating coil and the heating coil provided in the supply pipe. 5. The method according to claim 4, wherein the checking characteristic includes at least one of a predetermined water flow rate flowing through the heating coil, a feed air flow rate flowing into the room, an air temperature flowing into the room, an air temperature flowing into the heating coil, The temperature of the incoming water and the feed-forward control signal. 6. The method of claim 5, wherein the verification feature further comprises an authority and a percentage of a water flow rate through the valve, wherein the authorization is based on a change in hydraulic pressure across the valve from a circuit disposed with the valve fully open Wherein the ratio is a ratio to a change in pressure of the indoor unit. 7. The method of claim 6, wherein the verification characteristic is indicative of a thermal measurement supplied to the adjacent air by the heating coil, R = (T _{a, o} -T _{a, i} ) / (T _{f, i} -T _{a, i} ) (here T _{f, i} Lt; RTI ID = 0.0 > T _{a, i} The air inlet temperature, T _{a, o} The air discharge temperature) And a coil efficiency variable R, which is defined by the following equation: < EMI ID = 1.0 > 6. The method of claim 5, wherein the water flow rate T _{f, o} And < RTI ID = 0.0 > (Where K is a constant and ratio for the mass capacitance generation of air and water) And the water flow rate is calculated by using the temperature sensor. An air supply line, at least one general air exhaust line, and a room temperature regulator for regulating the temperature to a pre-set value and controlling the air inlet and outlet to the room to maintain a pre- (HVAC) fluid distribution system that affects at least one of an air temperature and an air pressure in a specific room, Confirmation means for periodically generating verification characteristics of the component to be controlled, Wherein the thermal energy flowing into the room is substantially the same as the thermal energy discharged from the room and the mass of the air flowing into the room is substantially equal to the mass of the air discharged from the room, Forwarding means coupled with verification means for generating a rope forward control signal, A feedback means for generating a feedback signal based on the measured system parameter, And means for combining the feed forward control signal and the feed back control signal for control of the local component. 10. The indoor temperature control apparatus according to claim 9, wherein the control set value includes a flow rate set value of the supply air, a discharge temperature set value of the supply air, and a flow rate setting value of the general exhaust air. 11. The indoor temperature control device according to claim 10, wherein the control set value is changed suitably as a function of the change of the confirmation characteristic. 12. The indoor temperature control apparatus of claim 11, wherein the component comprises a supply air damper / actuator, a general exhaust damper / actuator, a heating coil, and a valve / actuator. 13. The indoor temperature control apparatus according to claim 12, wherein the checking characteristic comprises an air flow rate flowing into the room, a power supply damper / actuator power, and a temperature setting value. 13. The room temperature regulating apparatus according to claim 12, wherein the checking characteristic comprises a flow rate of the exhaust air flowing out of the room, an authority of the discharging damper / actuator, and a temperature set value. 13. The indoor temperature control device according to claim 12, wherein the component comprises a heating coil installed in a supply pipe and a valve / actuator for controlling a flow of water flowing through the heating coil. 16. The method as claimed in claim 15, wherein the checking characteristic comprises at least one of a predetermined water flow rate flowing through the heating coil, a feed air flow rate flowing into the room, an air temperature flowing into the room, an air temperature flowing into the heating coil, The temperature of the incoming water and the feed-forward control signal. 17. The method of claim 16, wherein the verification characteristic further comprises an authority and a percentage of a water flow rate through the valve, wherein the authorization is based on a change in hydraulic pressure across the valve, Wherein the ratio is a ratio to a change in pressure of the indoor unit. 12. The method of claim 11, wherein the supply air flow rate set point ( ) And supply air temperature set point ( T _{s | sp} ), ? P _{| sp} = _{pref | sp} -p _{| sp} And (here P _{s | sp} A supply air pressure set value, T _{s | sp} A supply air temperature set value, Lt; RTI ID = 0.0 > a < / RTI & P _{ad | sp} The infiltration air pressure set value, T _{ad | sp} The infiltration air temperature set value, The infiltration flow rate set value, P _{| sp} The indoor pressure set value, An exhaust flow rate set value, T _{| sp} The indoor temperature set value, C _f Is a unit conversion factor) And the temperature of the room is controlled by the control unit. 12. The method of claim 11, wherein the exhaust air flow rate set point ( 0.0 > 0 < / RTI > when heating of the feed air is required and no heating of the feed air is required, (here ? P _{| sp} = _{pref | sp} -p _{| sp} And ego, P _{s | sp} A supply air pressure set value, T _{s | sp} A supply air temperature set value, Lt; RTI ID = 0.0 > a < / RTI & P _{ad | sp} The infiltration air pressure set value, T _{ad | sp} The infiltration air temperature set value, The infiltration flow rate set value, P _{| sp} The indoor pressure set value, Is a discharge flow rate set value, T _{| sp} Room temperature set value) And the temperature of the room is controlled by the control unit. 17. The indoor thermostat of claim 16, wherein the verification characteristics of the heating coil, the valve / actuator and the damper are determined by a general regressive neural network (GRNN). 21. The method of claim 20, And the temperature of the room is controlled by the temperature sensor. A method for determining the flow rate of water flowing into the room through a supply pipe installed in a supply pipe in a room having a pipe including an exhaust pipe, a pipe for measuring the air temperature in the pipe and an outlet for measuring the temperature of the coil, Temperature of the water at the outlet of the coil T _{f, o} , ≪ / RTI > Air temperature in supply air line T _{a, i} , ≪ / RTI > Air temperature in the exhaust air pipe T _{a, o} ; And Da food (Where K is the constant and ratio of the mass capacitances of air and water) Water flow rate ≪ / RTI > ≪ / RTI > The air supply pipe is part of the HVAC system of the building and the air supply pipe is provided with a heating coil for heating the air to be transferred through the air supply pipe and a flow valve for controlling the flow of hot water flowing through the heating coil, Comprises a check means for periodically generating the checking characteristics of the heating coil and the valve, an air flow rate measuring means for passing through the air flow pipe, and means for measuring the water pressure across the valve in the HVAC system to which the valve is connected The control signal is based on the control set value and the checking characteristics of the coil and the valve. Thus, a method for determining the value of the control signal in the temperature control device for controlling the temperature of the outlet side of the air to be introduced into the room from the air supply pipe, Activating the verification means to determine a heat exchange rate of the coil in the heat transferred to the air delivered through the air supply line; Using the coil characteristic to calculate a predetermined water flow rate flowing through the heating coil with respect to a given air temperature and air flow rate at a measurement air supply line outlet; Determining a ratio of the pressure drop across the valve to the pressure drop across the system to estimate the pressure drop across the valve to the total pressure drop of the system when the valve is fully open and to infer the right value of the valve step; And generating the control signal as a function of the water flow rate and the valve authority. The heat exchanger according to claim 23, wherein a temperature of the incoming water to the heating coil, an inlet air temperature of the upstream air supply pipe of the heating coil, and an outlet air temperature of the downstream side of the heating coil are measured, R = (T _{a, o -} T _{a, i} ) / (T _{f, i -} T _{a, i} ) (here T _{f, i} The incoming water temperature T _{a, i} Lt; RTI ID = 0.0 > T _{a, o} The exhaust air temperature) ≪ / RTI > further comprising the step of: In the interior of a building having a heating, ventilating and air conditioning system, the room has an air supply pipe for supplying air to the room and at least one exhaust pipe for discharging air from the room. A method for determining a thermal load of a room so that the air pressure in the room is relatively constant, Determining a room temperature using a temperature sensor; Determining a feed flow rate at a preceding time t-1; Determining an exhaust flow rate from the room at an annual time step t; Determining a flow velocity of infiltrated air; And (here ego, Is a set value of infiltration flow rate, (here ? P _{| sp} = _{pref | sp} -p _{| sp} ) ≪ / RTI > The method comprising the steps of: 26. The method of claim 25, wherein the temperature sensor is installed in an exhaust pipe of the room.